The Milankovitch theory states that the orbital eccentricity, precession, and obliquity of the Earth influence our climate by modulating the summer insolation at high latitudes in the northern hemisphere. Despite considerable success of this theory in explaining climate change over the Pleistocene epoch (2.6–0.01 Myr ago), it is inconclusive with regard to which combination of orbital elements paced the 100 kyr glacial–interglacial cycles over the late Pleistocene. Here we explore the role of the orbital elements in pacing the Pleistocene deglaciations by modeling ice-volume variations in a Bayesian approach. When comparing models, this approach takes into account the uncertainties in the data as well as the different degrees of model complexity. We find that the Earth's obliquity (axial tilt) plays a dominant role in pacing the glacial cycles over the whole Pleistocene, while precession only becomes important in pacing major deglaciations after the transition of the dominant period from 41 kyr to 100 kyr (the mid-Pleistocene transition). We also find that geomagnetic field and orbital inclination variations are unlikely to have paced the Pleistocene deglaciations. We estimate that the mid-Pleistocene transition took place over a 220 kyr interval centered on a time 715 kyr ago, although the data permit a range of 600–1000 kyr. This transition, occurring within just two 100 kyr cycles, indicates a relatively rapid change in the climate response to insolation.